After having described the recent trends in health status, it is necessary to identify its determinants. Knowing the determinants of health status, of both objective health status and its subjective perception, allows the design of a more targeted policy and thus the opportunity to cut costs. In this section a multivariate analysis is undertaken to identify the determinants of health status using two different measures: a subjective indicator, which is satisfaction with health, and a more objective one, which is the degree to which an individual is hampered when fulfilling daily activities. The descriptive analysis above has shown that the previous trends for these indicators moved in opposite directions – while the subjective perception of personal health status worsened during 1984-2000 and only recently improved, the individuals observed in each time period were less hindered in their daily life, which indicates an improvement in the general health status of the population. Knowledge about the determinants of the two measures and their comparison may identify those who have increasing expectations and those who really are in a worse health condition.
Using a sample of roughly 21,000 people,4 an ordinary least square regression has been run controlling for the impact of the following variables: personal variables such as age, age squared, gender and education (measured in years of schooling); marital status (five dummies for married, which is the omitted group, single, separated, divorced or widowed); and origin (which can be either West or East Germany). Furthermore, socio-economic variables have been used such as job, described by its prestige (using the Wegener scale of the magnitude of job prestige from 0 to 200) and income after subsidies and taxes, along with the type of health insurance (public or private). Dummies for the years, grouped into five years, and a variable indicating the cohort (1930-70) have also been included. The dependent variable of satisfaction with health has been scaled from 1 (very unsatisfied) to 10 (very satisfied). The degree to
4 The size of this sample is significantly smaller than those presented in the first section. This is owing to missing values for certain variables, especially income, job prestige and type of insurance.
OBJECTIVE TRENDS AND PERCEPTIONS OF HEALTH STATUS IN GERMANY |11
which one is hindered in fulfilling daily life activities ranges from 1 (not hindered at all) to 3 (completely hindered). The results can be viewed in Table 2.
Table 2. Regression results for health status
Health satisfaction Hindered by health
Age -0.07 -0.07
(-11.26)** (5.43)**
Age squared 0.00 0.00
(8.80)** (-0.82)
Female -0.15 -0.15
(-10.16)** (-9.08)**
West Germany 0.07 0.07
(3.69)** (1.59)
Education (in years) 0.02 0.02
(3.82)** (8.74)**
Separated 0.19 0.19
(3.44)** (5.11)**
Divorced 0.00 0.00
(0.09) (0.26)
Widowed 0.10 0.10
(1.95)* (2.06)
Single -0.02 -0.02
(-0.67) (1.72)
Children 0.01 0.01
(0.67) (1.25)
Ln job prestige 0.05 0.05
(2.09)* (2.27)**
Ln income 0.09 0.09
(5.55)** (4.98)**
Private insurance 0.02 0.02
(0.85) (2.13)**
Cohort 0.00 0.00
(-0.29) (0.69)
Year 1990-94 -0.07 -0.07
(-2.04)* (-1.67)
Year 1995-99 -0.14 -0.14
(-3.02)** (-1.07)
Year 2000-03 -0.13 -0.13
(-2.21)** (-1.01)
Note: The absolute value of z statistics are in parentheses; *significant at 5%; ** significant at 1%
Source: Author’s calculations based on the GSOEP (1984-2003).
As the bivariate descriptive analysis has clearly indicated, health status worsens for older ages.
Both the level of satisfaction with health and the degree of being hindered seem to grow worse with old age. Yet, while the negative effect on health satisfaction becomes smoother at older ages, the degree of handicap rises monotonically.
The multivariate analysis gives a better idea about the impact of gender on health status.
Although women are generally less satisfied with their health status as observed in the descriptive analysis above, being female also has a significant impact on the degree of being hindered. Furthermore, geographical origin seems to have an effect on health status, significantly so with regard to perception. Citizens in West Germany seem to be more satisfied with their health status. Nevertheless, origin does not have a significant effect on the degree of being hindered. This result is interesting, as it seems that East Germans perceive their health status to be worse even if their daily life is not more affected by their health condition than that of West Germans. The result of the coefficient for marital status is surprising, given that separated and widowed persons seem to be in better physical shape than their married counterparts. It could be the case that they have fewer family duties and care more about their physical conditions.
Economic variables such as human capital (represented by education) and a better job with respect to both its prestige and its income seem to have a significantly positive impact. A higher level of education and a better job could influence health status in several different ways. First, well-educated individuals may be more conscientious about their health and thus take more care, as well as undergo preventative medical check-ups more often. In a decomposition of the probability of receiving medical care with respect to education, it can be confirmed that individuals with higher education (especially at younger ages) visit a doctor more frequently.
Second, a better job may be less physically demanding and thus have less severe physical consequences. Disentangling the effect of job quality shows that occupations associated with more prestige seem to have fewer negative effects on the degree of disability. Individuals with a better job are also less likely to stay in a hospital, which may stem from fewer accidents at work. Finally, another way in which a better job may affect health status is the amount of income it brings. A higher salary may enable people to afford better health services, in terms of both prevention and care. Individuals with higher income (above €3,700 gross income per month) can opt out of the statutory health insurance system and participate in a private plan.
Private health insurance facilitates access to higher quality health care services and pays for more treatments, health care products and medications. In the regression above we can see that the presence of private insurance seems to significantly affect the degree to which a person is hindered by his/her health. Yet while the privately insured seem to be less hindered by their health status they do not seem to perceive a better health condition. This finding could indicate that these individuals have higher expectations about their health.
Before concluding, it is important to look at the time effect, in terms of both years and cohorts.5 The coefficients of the different year groups studied show the same pattern already observed in the bivariate analysis: health status has worsened over the last two decades and only recently do people perceive it to be better even if it did not improve significantly. Interestingly, the cohort does not seem to be important with regard to an objective assessment of health status but has a
5 In a regression where year dummies and an indicator of the cohort are included, the effects are not significant. Separate regressions, however, in which one of these two is omitted, show significant effects.
OBJECTIVE TRENDS AND PERCEPTIONS OF HEALTH STATUS IN GERMANY |13
negative impact on its perception. These results indicate that younger cohorts have higher expectations and thus believe their health to be worse even in the absence of objective indicators.